Regional End-to-End Disaster Pilot
Project in Namibia
for GEOSS AIP-5 Kick-off meeting, Geneva Switzerland
Stuart Frye – NASA/GSFC
May 3, 2012
1
Where is Namibia
2
2
Stakeholders
• Namibia Department of Hydrology
• University of Namibia
- Department of Geography
- Multidisciplinary Research Center
-
Namibia Ministry of Health
• NASA/GSFC
• University of Maryland, Department of Geography
• University of Oklahoma
• University of Chicago
• Open Cloud Consortium
• Committee on Earth Observing Satellites (CEOS)
• Disaster Societal Benefit Area
• Working Group on Information Systems and Services (WGISS)
3
Project Objectives
•
Define & facilitate implementation of a sensor web-based architecture for
risk management from a multi-hazard perspective
•
Address scope of GEOSS Task (Disaster Management DI-01-C5)
•
Expected Impact:
•
•
Reduce the time to acquire and improve the use of relevant satellite data for
flood assessment and forecasting
•
Increase the usefulness of derived satellite flood data products for local
populations
Approach:
•
Coordinate with WGISS to document and prototype a disaster management
architecture to demonstrate improved decision support capability and access to
remote sensing assets
•
Conduct socioeconomic surveys in flood prone areas
• Identify local concerns/cultural barriers which prevent use of local flood forecasts
• Explore methods to incorporate local observations into decision support systems and
social networking technology (e.g., crowd sourcing)
4
Objectives Illustrated
5
NASA Flood SensorWeb Concept
Trigger
automatic high
resolution
satellite data
acquisition
Task Multiple Sensors in
Area of Interest
Detect Heavy Rains and Floods
Upstream at a coarse level
(River Gauge)
Acquire
Satellite Data
(Images)
Analyze Risks
Automatically
Acquire Insitu Data
(1) Automatically update model and
validate (2) Display data on Web
6
Analyze Images
SensorWeb High Level Architecture
Sensors, Algorithms and Models Wrapped in Web Services Provide Easy Access to
Sensor Data and Sensor Data Products
floods, fires,
volcanoes etc
Data Processing Web Services
Node
Geolocation,
SWE Node
Level 2
Get satellite
images
RSS Feeds
Sensor
Data Products
Internet
OpenID 2.0
GeoBPMS
SWE Node
UAV Sensor Data Node
Orthorec,
algorithms
Coregistration,
(e.g.
atmospheric
flood
extent)
correction
In-situ Sensor Data Node
Level 0 and
Level 1
processing
SWE Node
EO-1
Satellite
SAS
Web Notification
Service SOS
(WNS)
Sensor Planning
WFS
L1G Service
(SPS)
Sensor
Observation
SPS
Service (SOS)
Satellite Data Node
Web Coverage
Processing
Service (WCPS)
Workflows
Task satellites to provide images
7
Design new
algorithms and
load into cloud
Cloud Configuration for Flood Dashboard
CREST Hydrological
Model
Namibian River
Gauge Stations Daily Measurements
TRMM based Global
Rainfall Estimates
Radarsat Images &
flood extent maps
MODIS Daily Flood
Extent Map
Global Disaster
and Alert and
Coordination
System (GDACS)
•Eucalyptus/Open Stack-based Elastic
Cloud SW
•300+ core processors
•40 x 2 Tbytes of storage
•10 Gbps connection to GSFC
- being upgraded to 80 Gbps (Part
of OCC)
•Hadoop/Tiling
•Supplied by Open Cloud Consortium
•Open Science Data Cloud Virtual
Machines & HTTP server to VM’s
Storage – 1 year
Hyperion & ALI Level 1R
Storage – 1 year
Hyperion & ALI Level 1G
Storage – 1 years
Hyperion & ALI Level 1R
and Level 1G AC
Storage – 1 year
User Defined L2
Products
e.g. EO-1 Flood Mask
Namibia River Gauge
Data base
Flood Dashboard
Display Service
- Mashup
- Google Maps Inset
- Plot Package
http server
8
Hadoop and Tiling Handles Large Dataset
Displays
Storage – 1 year
Hyperion & ALI Level 1R
Storage – 1 year
Hyperion & ALI Level 1G
Storage – 1 years
Hyperion & ALI Level 1R
and Level 1G AC
Storage – 1 year
User Defined L2
Products
e.g. EO-1 Flood Mask
Hadoop / HBase
Partition into Cloud Cache
Suitable for
Google Earth / Open Layers
Web map
Service
(WMS)
9
HBase storage
of multiple
missions over
multiple days
Google Earth View of High Population and
High Flood Risk Area in Northern Namibia
Angola
Objective
Namibia
Shanalumono
River Gauge Station
Key
Milestones
Approach
Co-Is/Partners
10
EO-1 Satellite Image of High Risk Flood Area
in Northern Namibia
Shanalumono river
gauge station taken
from helicopter
Earth Observing 1 (EO-1)Advanced Land Image (ALI)
Pan sharpened to 10 meter resolution, Oshakati area
11
TRMM based rain
estimates. Monitor
rains in Northern
basins that drain
Objective
into Namibia.
Global Disaster and
Coordination
System- (Based on
AMSR-E)
Early user alert
Shanalumono
River Gauge Station
GeoBPMS
(Tasks satellites
in an area of
interest)
Approach
Daily flood
gauge levels &
predicted river
levels plots
Auto-trigger
Hi-res Satellite imagesKey
MODIS Daily Flood
Mask
Follow flood wave
down basin
Milestones
Select river gauges, coarse daily flood
extent maps or models/pre-warning
probability to auto trigger high
resolution satellite data acquisition
High resolution
satellite imagery (e.g.
Co-Is/Partners
EO-1)
Flood
Dashboard
(mashup)
CREST Model
and prewarning
probability
High Level Diagram of Namibia Flood SensorWeb
for Early Warning
12
Training for Data Capture
Georeferenced photos to enable Rob Sohlberg from
Univ. of Maryland to train classifier to detect presence of
water in grassy marsh lands via from satellite data.
McCloud Katjizeu (orange) Dept of Hydrology compares
GPS readings of control point with UNAM students for
mapping exercise
13
Flood Impact on Wild Life and Subsequently
on Humans Nearby
Hippo tracks near villager
Crop fields. Hippo crop
destruction is big impact to villagers.
14
Request from Namibia Hydrological Services
• Monitor flooding in near-real time
• Create classification products
– partition floodwaters by turbidity
– presence of grasses, etc.
• Demonstrate rapid prototyping utilizing Web Coverage Process
Services (WCPS)
– To be used to both inform civil managers and – more importantly –
to developed and validate predictive models.
• Improved hydrological model based on CREST
– Model developed by University of Oklahoma
• Improved data products pipeline ( more automation)
15
15
Contributions of Namibian Partners
•
•
•
•
•
Local terrain expertise to improve modeling
Knowledge of local populations
Expertise on conducting local surveys
Develop new techniques and products useful to decision makers
Namibian model will be extended to other countries and applications.
Dan Mandl/NASA, Alphons Mosimane/UNAM, Selma Lendevlo//UNAM,
Dr. Julie Silva/UMD, Dan Mandl/NASA, Victoria Shifidi/Dept Hydrology ,
Dr. Simon Angombe/UNAM, Margaret Angula/UNAM
Hydrology team begins river
validation exercise
Socioeconomic team discusses desired outcomes, timeline and next
steps to develop a village level study which is integrated with
Hydrology Dept. effort.
16
Planned Technical Enhancements to Flood Dashboard
Current Flood Dashboard on Computation Cloud
17
17
Planned Technical Enhancements to Flood Dashboard
•
•
•
•
•
•
Partnership with Canadian Space Agency to simultaneously task EO-1
and Radarsat and deliver products from both automatically to Flood
Dashbaoad
Color code river stations to indicate underlying states and click for
details
More options on hydrographs
– Min
– Max
– Average
– Rainfall plots overlay
Daily excel file download from FTP site and which provides underlying
status and color codes of river gauges
Track visitors to site
Prototype pre-early flood warning which shows probability of flooding for
next two weeks (via a plot) and is updated daily by model (see next
slide)
–
Developed by University of Oklahoma (Yang Hong and Zac Flamig)
18
18
Planned Technical Enhancements to Flood Dashboard
•
Obtained data from Central Bureau of statistics on this trip. Put on
dashboard
– New dwelling unit data base with schools, roads, commercial buildings,
hospitals etc. geolocated and identified by class
– Begin to evaluate how to use flood data with this data base (e.g. roads
blocked due to floods)
•
•
Collected hundreds of GPS encoded photos to enhance /calibrate
flood classification algorithm for EO-1
Met with Ministry of Health, National Vector Borne Disease Control
Program, personnel to discuss adding water-borne disease risk maps
as layer on Flood Dashboard
– Dr. Petrina Uusikia, Chief Medical Officer, National Vector-borne Disease
Control Program, Ministry of Health
– Closhilde Narib,
•
Met with Dr. Martin Hipondoka, head of Remote Sensing at Univ. of
Namibia
–
Trying to arrange training on classification for floods using EO-1 optical imagery and
Radarsat imagery
19
19
Contact Information and Background
•
•
•
•
•
•
•
•
stuart.frye@nasa.gov
The following two pages contain a short list of the servers we have setup that
are based on Open Geospatial Consortium standard interfaces.
We have not implemented all the OGC standards, but you can find examples
of the rest by checking the OGC website http://www.opengeospatial.org
The idea is that services and components are on the internet, so you can
access them with http:// or https://
Servers should have an XML description document that tells clients how to
access things that server has
The servers should have an Application Programmer Interface so clients can
automatically access them via scripted actions, not just manual browser
actions
The servers should have enough security to ensure that the users are who
they say they are so actions taken by the clients can receive delegated
authority to do the same things that the user who launched the client are
allowed to do
Each of these have a long list of needed improvements, but the rudimentary
ideas are demonstrated OK
20
URL’s and Descriptions
• OpenID Provider-Server = https://op.geobliki.com/ controls the
security (this is where you setup your account)
• Campaign Manager = http://geobpms.geobliki.com/home allows
tasking requests to be submitted (i.e., targets)
• EO-1 Server = http://eo1.geobliki.com/ this is where EO-1 data
can be found along with the status of future and past taskings
• Radarsat Server = http://radarsat.geobliki.com/radarsat where
we provide access to Radar raw data, browse images,
metadata, and processed flood produicts
• MODIS Flood Server (API) = http://modis.geobliki.com/modis is
where you can point your browser to manually check on daily
MODIS flood maps
• MODIS Flood Server (GUI)
= http://oas.gsfc.nasa.gov/floodmap/ is the server that provides
and API for accessing the daily MODIS maps
21
URL’s Concluded
• Flood Dashboard Client =
http://matsu.opencloudconsortium.org/namibiaflood this is an
example of a client implementation that runs on a cloud computing
platform provided through a collaboration with the University of
Illinois/Chicago
• WCPS Server =
http://matsu.opencloudconsortium.org/wcps/session/login is where
you go to generate and run algorithms against satellite data
• Pub/Sub Server = http://opsb.geobliki.com/session/new is where
you setup a subscription for requesting notifications about new data
in your area or from a particular instrument or with a particular
feature or….The notifications come via Email, SMS, or twitter and
contain RSS or Atom feeds for you to follow to find the processed
or raw data. Clients can be automated to monitor the feeds and
pull the data they are programmed to look for
22